Lipstick Image Classification Dataset

The current retail e-commerce industry faces challenges in product recognition and recommendation system accuracy. Many existing image classification techniques perform poorly with diverse products and complex backgrounds, leading to decreased user experience. To enhance lipstick product recognition and recommendation effectiveness, this dataset aims to provide high-quality lipstick image data to solve technical difficulties in image classification. The dataset includes 5000 lipstick images, taken with professional equipment under good lighting conditions to ensure image quality. For quality control, we have implemented multiple rounds of annotation and expert review to ensure consistency and accuracy in the annotations of each image. The data is stored in JPG format, organized by associating image files with their metadata. The core advantage of this dataset is its high data quality, with annotation accuracy exceeding 95% and consistency reaching 90%. New annotation methods and data augmentation techniques have significantly improved the model's classification performance. Validation results show that under the same conditions, models using this dataset have improved accuracy by 15%.